Earth Systems and Environment (2020) 4:225–244 https://doi.org/10.1007/s41748-019-00141-w

ORIGINAL ARTICLE

Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow Model

Md Abdullah Al Baky1 · Muktarun Islam2 · Supria Paul3

Received: 14 August 2019 / Accepted: 20 November 2019 / Published online: 5 December 2019 © The Author(s) 2019

Abstract This study is concerned with food risk that can be assessed by integrating GIS, hydraulic modelling and required feld information. A critical point in food risk assessment is that while food hazard is the same for a given area in terms of inten- sity, the risk could be diferent depending on a set of conditions (food vulnerability). Clearly, risk is a function of hazard and vulnerability. This study aims to introducing a new approach of assessing food risk, which successfully addresses this above-mentioned critical issue. The food risk was assessed from food hazard and vulnerability indices. Two-dimensional food fow simulation was performed with Delft3D model to compute foodplain inundation depths for hazard assessment. For the purpose of food vulnerability assessment, elements at risk and food damage functions were identifed and assessed, respectively. Then, fnally food risk was assessed frst by combining replacement values assessed for the elements and then using the depth–damage function. Applying this approach, the study fnds that areas with diferent levels of food risk do not always increase with the increase in return period of food. However, inundated areas with diferent levels of food depth always increase with the increase in return period of food. The approach for food risk assessment adopted in this study successfully addresses the critical point in food risk study, where food risk can be varied even after there is no change in food hazard intensity.

Keywords Flood · Hazard · Vulnerability · Risk · Hydraulic model

1 Introduction is a deltaic country located at the lower part of the basins of the three mighty rivers—the , the This study is concerned with how we can assess food risk in Brahmaputra and the Meghna. This unique geographical a given food prone area. There are many examples of food setting, surrounded by mountains on three sides, together studies in diferent countries. Bangladesh is one of those with extremely fat and low-lying foodplain topography, a countries where a signifcant number of studies have been low-lying coastline, and an extreme climate variability has carried out with food issue. This research selected Bangla- rendered the country highly prone to natural hazards such desh for the purpose of assessing food risk. as food (Chowdhury et al. 1997; Hoque et al. 2011; Islam et al. 2010). About one-ffth to one-third of the country is annually fooded by overfowing rivers during pre- * Md Abdullah Al Baky (April to May) and monsoon (June to September) periods. [email protected] These foods cause physical damages to agricultural crops, Muktarun Islam buildings and other infrastructure, social disruptions in vul- [email protected] nerable groups, livelihoods and local institutions, and direct Supria Paul and indirect economic losses (Baky et al. 2012; Bhuiyan [email protected] and Dutta 2012; Mirza 2011). The food hazard problem in recent times is getting more and more frequent and acute due 1 School of Geography, University of Melbourne, Melbourne, Australia to growing population size and human interventions/socio- economic activities in the foodplain at an ever-increasing 2 Agricultural University, Sylhet, Bangladesh scale (Bhuiyan and Baky 2014; Paul and Routray 2010). 3 Department of Geosciences, University of Rhode Island, South Kingstown, USA

Vol.:(0123456789)1 3 226 M. A. A. Baky et al.

Flood mitigation approaches in Bangladesh have ranged of storm surge prone area of the Ganges tidal plain. Haz- from structural interventions (embankment, fow regulation ard factors were based on simulated spatial distribution structures, etc.) to non-structural approaches (forecasting of 100-year food depths, while the vulnerability factors and warning, food preparedness before during and after the were based on the distribution of population densities. food, food proofng measures, etc.) (Paul and Routray 2010; Risk indices were then derived as a product of hazard Rahman and Salehin 2013). The Flood Action Plan (FAP) and vulnerability factors as per Skakun et al. (2014). At and French Engineering Consortium (FEC) conducted a that time, the application of hydraulic simulation and or feasibility food control survey and a hydrological study in satellite image in preparing food hazard map was not so Bangladesh. They focused on agricultural adjustment pro- prominent. Therefore, the study outputs had question with cesses (Younus 2012) and the impacts of foods when food accuracy. Afterwards, Islam and Sado (2000) used NOAA- adaptation fails at the community level (Younus and Harvey AVHRR images with GIS to develop food hazard maps 2014). Structural food protecting projects even though have for Bangladesh. Tingsanchali and Karim (2005) studied short-term positive impacts, the viability of those projects food hazard, vulnerability and risk in the southwest region made over the years has often faced criticism because of of Bangladesh. However, problem was that they assumed their adverse hydraulic, environmental and socio-economic the vulnerability factor to be proportional to population impacts (Burrel et al. 2007; Chowdhury et al. 1997; Shaw density, which does not truly represent the food damage 2006). Those projects cause economic hardship to the poorer data, thus questioning the output of vulnerability as well community of the society, which are dependent on many free as risk. For assessing vulnerability, at frst the approach resources (e.g. fsheries) of river (Chowdhury et al. 1997; should address the elements at risk, and then estimates the Chowdhury 2010) and even on water transport by country damage function. This is the theoretical basis of assess- boats (Hunting 1992). Furthermore, the benefts of struc- ing vulnerability as given by Merz et al. (2007). Hasan tural food control projects cannot be reached at all level (2006) followed this theoretical basis for assessing agri- of society. For example, aquaculture activities are becom- cultural food vulnerability in Tarapur union (the smallest ing popular particularly in the coastal region of Bangladesh rural administrative unit of Bangladesh), Gaibandha dis- (Sohel and Ullah 2012) as polder-based food control project trict, Bangladesh. Masood and Takeuchi (2012a), Dewan is developing in the region. However, it is still unlikely for et al. (2007) and Dewan (2013) conducted food hazard the marginal population to engage in such aquaculture activi- and risk assessment for Dhaka, Bangladesh. These studies, ties since the activities need sufcient capital to start. particularly the study by Masood and Takeuchi (2012a), The experiences with structural food control interven- indicated to identify elements at risk by following the land tions gave way to new insights, which are a combination use mapping approach, thus improving the output of food of structural and non-structural food hazard mitigation vulnerability. measures depending on the specifc local or regional needs Recently, we can observe some milestones in food haz- (Van Alphen and Lodder 2006). Very recently, Bangla- ard and vulnerability studies, but not in food risk study. desh government has focused on both structural and non- The application of hydraulic model in food hazard assess- structural food management approaches to reduce vulner- ment is becoming popular. We can mention the studies by ability to fooding in the country (Paul 1995, 1997; Paul Aff et al. (2019), Zin et al. (2018), and Tyrna et al. (2018) and Hossain 2013). Preparation of food risk maps is the that applied diferent hydraulic models in studying food basic requirement before non-structural food hazard miti- hazard. At the same time, land use-based and community- gation approaches (Bhuiyan and Baky 2014; Bhuiyan et al. based food vulnerability study is also getting prioritised. 2014; Demir and Kisi 2016; Giustarini et al. 2015; Hoque The study by Masood and Takeuchi (2012b) is a comprehen- et al. 2011; Islam et al. 2010; Islam and Sado 2000). Flood sive study showing a clear path of how to conduct food risk risk mapping facilitates the administrators and planners to study. The study shows the uses of one-dimensional hydrau- identify areas vulnerable to food hazard and to determine lic model for hazard assessment and analysis of land use infrastructure at risk and the degree they might be afected, for vulnerability assessment and fnally fnds out food risk and to map their capacity to respond and recover (Bhuiyan from hazard and vulnerability. Rakib et al. (2017) and very and Baky 2014; Bhuiyan et al. 2014; Hazarika et al. 2018; recently Hoque et al. (2019) introduce a new dimension in Sanyal and Lu 2009; Tran et al. 2009; Vojtek and Vojteková vulnerability assessment, and this dimension makes the vul- 2016). Most importantly, it helps in identifying and prioritiz- nerability assessment become more community focused than ing the mitigation and response eforts and helps to inform before. Their methods include analysing satellite images, a emergency responses (Tran et al. 2009). structured questionnaire, criteria mapping, observation and Despite having mentionable positive options of food secondary data. risk map, the study of flood risk mapping is limited. Overall, the studies of food hazard and vulnerability are Chowdhury and Karim (1996) studied on risk-based zoning good in number as well as they are updated. However, still

1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 227 now food risk as a concept has not been addressed prop- the (wetland) areas in the northeast region (Brammer erly in food studies. Flood risk is a function of hazard and 1999). They cause massive damage to dry-season boro (rice vulnerability (Alexander 1991; Skakun et al. 2014). It is the variety) rice crop just before or at the time of harvesting. In probability of loss due to food of a given intensity (Alex- case of property loss, they cause breaching to embankment ander 1991). The critical point is that while food hazard is and other food controlling structures, road, railway, bridges, the same for a given area in terms of intensity, the risk could buildings, etc. In this study, one of such fash food hazard be diferent depending on a set of conditions and this set prone areas, the Baniachong (sub-district) in Hab- of conditions is referred by vulnerability (Crichton 1999). iganj district, was selected for assessments of food hazard, Since knowledge on food risk is must for a planner to plan a vulnerability and risk (Fig. 1). disaster resilience society, addressing food risk properly is a with an area of 482.25 km2 is prerequisite, but it has not done yet. This study aims to intro- located in the Agro-ecological Zone—20: Eastern Surma- duce a new approach of assessing food risk properly, which Kushiyara Floodplain. The mean annual rainfall in the area successfully addresses this above-mentioned critical issue. is about 2659 mm, as compared to the national average of The focus of the food risk of this study is the Baniachong 2300 mm. The rainfall varies considerably within a year, Upazila (Sub-district), one of the food-afected with 79% of rainfall occurring in 5 months from May to in Bangladesh. Adopting the hydrodynamic behaviour cer- September. tainly increases the accuracy of food hazard map (Aff et al. The river network close to Baniachong includes Shutki, 2019), thereby ensuring better accuracy in food vulnerabil- Old Kushiyara, Shaka, Borak, Shingli, and Bibiyana rivers ity and risk maps. This study applied a 2D hydrodynamic (Fig. 1). These rivers are hydraulically connected with the model for the purpose of producing food hazard map at dif- Kushiyara–Kalni river system which fows down along the ferent return periods of foods in the study site. The hazard north of the sub-basin from northeast towards the southwest maps were assessed under diferent land use categorizations. and provides a major source of food water during the mon- The subsequent attempt was to prepare food vulnerability soon. The major portion of the basin lies outside Bangladesh maps from food depth–damage functions for cropping land and the area receives water from Hills on the south and rural settlements. As a fnal attempt, the study estimated and the run-of from right bank foodplain of the Kushi- damage and produced risk maps from vulnerability index yara–Kalni. The topography of the foodplain in the study and hazard index. area is generally fat with some depressions located in north and southwest portions of the study area (Elevation map— Fig. 1). The area surrounding the CBD (Central Business 2 Methodology District) of the Upazila is positioned at higher elevation than distal foodplain of the local rivers (Elevation map—Fig. 1). An integrated and interdisciplinary research approach was followed in this study. Technical assessments were inte- 2.2 Derivation of Design Floods grated with stakeholders’ views on diferent aspects of food hazards and risks. The interdisciplinary nature of the present 2.2.1 Selection of Gauge Stations study warranted the use of a wide range of technical and social research tools and methods. These are hydraulic mod- A number of water level gauge stations were selected on the elling (with GIS application) of food inundation, analysis main rivers surrounding the study area to be subsequently of satellite images to determine diferent physical elements used in determination of diferent food levels corresponding at risk and using quantitative method (questionnaire survey) to diferent return periods and subsequent analysis for food- to gather information about food damages to diferent types plain inundation. Figure 1 shows the locations of such gauge of physical risk elements in the study area. stations. The stations selected include at Sherpur (SW_175.5), River at Habigang (SW_159), 2.1 Description of the Study Area Surma-Meghna at Markuli (SW_270) and Surma-Meghna at Azmiriganj (SW_271). The principal sources of foods in Bangladesh are the river foods from the major river systems in the monsoon months. 2.2.2 Frequency Analysis of Water Levels A broad strip of land extending beyond the active river foodplains is subjected to this type of food. The northern For study area, frequency analysis of the maximum water and north-eastern trans-boundary hill streams are suscep- level and or discharge during the months of April–May tible to fash foods from the adjacent hills in in the in pre-monsoon season was conducted since it is the fash pre-monsoon months of April and May. Flash foods cause food inundation area. The source of the water level data is extensive damages to crop and property, particularly in Bangladesh Water Development Board (BWDB). BWDB

1 3 228 M. A. A. Baky et al.

Fig. 1 Location of Baniachong Upazila and selected gauge stations within the model boundary 1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 229 provides water level data at every 1-day interval. The time such as Delft3D, HEC-RAS, MIKE11, SOBEK, HEC-RAS, series water level and discharge data at the selected stations ISIS, ONDA and FLUCOMP is very common in food inun- were frst checked for trends. Frequency analysis was carried dation mapping at watershed level. For inundation model- out with diferent probability distributions functions (PDFs). ling, Delft3D, a 3D modelling technique developed in the These PDFs are: Two-Parameter Log Normal (LN2), Three- Dutch-based research institute (Deltares 2014) was used in Parameter Log Normal (LN3), Pearson Type III (P3), Log this study. Delft3D is a fully integrated computer software Pearson Type III (LP3) and Gumbel (EV1). To estimate suite applied to simulate hydrodynamics, sediment trans- peak discharge for diferent return periods, some of these port, waves, morphological developments, water quality and fve PDFs perform well in some cases. For example, Pum- ecology for fuvial, estuarine and coastal environments. The chawsaun (2018) found Log Pearson Type III (LP3) distri- Delft3D-FLOW, a module of Delft3D, is a hydrodynamic bution ft well in case of estimation of peak discharges for (and transport) simulation program which calculates non- 1-in-100 year food; on the other hand, Khan and Sabbir steady fow and transport phenomena resulting from hydrau- (2018) found that Pearson Type III (P3) is the best ftted lic and meteorological forcing on a curvilinear, boundary distribution for overall food frequency analysis. Since in ftted grid or spherical coordinates. this present study, it is unknown which distribution fts well An essential data required for Delft3D hydraulic simula- for the study area gauge stations; all these fve PDFs were tion is the land topographic data. The NASA Shuttle Radar compared to get the best ftted distribution. Table 1 shows Topographic Mission (SRTM) digital elevation data, avail- the equations of all these fve distributions. able as 1 arc second (approx. 30 m resolution), are very The PDFs were tested based on the probability plot cor- popular land topographic data. This dataset is found to use relation coefcient (PPCC) (Filliben 1975). Goodness-of-ft in many studies (e.g. Patro et al. (2009), Baugh et al. (2013)) test based on PPCC is useful for assessing whether a pro- related to food inundation mapping using fow model. In posed distribution is consistent with the at-site data sample this study, this SRTM data were used in hydraulic simula- (Stedinger 1993). The test uses the correlation coefcient tion. The DEM data were further processed using ArcGIS ‘r’ between the ordered observations and the corresponding to fll in the no-data voids or cells. The C-band radar signal ftted quantiles, determined by plotting positions for each of SRTM dataset is unable to penetrate through vegetation observation. Cunnane (1978) plotting position formula was canopy to the bare land surface, resulting in high absolute used to obtain the ftted quantiles. The best ft PDFs were vertical error in dense forest area (Baugh et al. 2013). How- selected and subsequently used to determine the design food ever, in open land surface this vertical error reduces signif- level. cantly (Rodriguez et al. 2006). Ahead of this problem, in this study, there was an attempt to process the SRTM dataset 2.3 Flood Inundation Mapping with vegetation height data to get more accurate output. The vegetation height dataset was collected from Spatial Data Preparing food inundation maps based on all primary and Access Tool (SDAT) site. The site provides 1 km Forest secondary data and hydraulic model is the post-stage of deri- Canopy Height globally. However, according to the dataset, vation of design foods. The use of hydrodynamic models canopy height was found zero in the study area. Therefore,

Table 1 Probability distribution Distribution Equation functions − log x−u2 Two-parameter log normal (LN2) f (x) = 1 exp y 2 2 x y 2 � y � log (x)−√ u = y y Three-parameter log normal (LN3) f (x) = 1 exp − 1 log (x − a) − 2 2 2 y (x−a) y 2 � y � √ −1 x− 1 x− − Pearson type III (P3) f (x) = e T()   log (x)− 1 log (x)− −1 − Log Pearson type III (LP3) f (x) = e xT() 

Gumbel (EV1) QT = b0 + b1YT Y =−ln ln T T T−1  

µy is the mean of the natural logarithms of x (sample variable); σy is the standard deviation of the natural logarithms of x; u is the standard normal variable; a is the shifted amount of x. QT is the fow corresponding to the return period T; and YT is the reduced variable

1 3 230 M. A. A. Baky et al. regarding the problem with vegetation height, no further processing was carried out with the SRTM dataset. The processed SRTM data were converted to the sizes required for 728 × 538 grids truncated for the study area using Delft3D-RGFGRID. Important considerations in con- structing the computational grids were: (i) the grids must ft as closely as possible to the land–water boundaries of the area to be modelled; (ii) the grids must be orthogonal, i.e. the grid lines must intersect perpendicularly; and (iii) the grid spacing must vary smoothly over the computational region. Figure 1 shows the model setup for the study area. Two boundary conditions were assigned: one upstream boundary and one downstream boundary. The upstream discharge was considered at location of Sherpur (SW175.5) which is part of the Kushiyara River and the downstream water level was considered at location of (SW 159) which is part of the . The model was simulated for return periods of 2.33, 10, 20, 50 and 100 years. Out of the available time series data at the boundary station, time series were selected such that the peak pre-monsoon water levels are close to the water levels analysed for diferent return periods at that station. Time step used in simulation for this model is 4 min and the simulation time spanned over 3 months from 1st Fig. 2 Workfow showing land use classifcation April to 31st May. land use and land cover classes: cropping land, rural settle- 2.4 Flood Damage Vulnerability Analysis ment, urban settlement, water bodies and bare land. Informa- tion collection during feld survey as ground-truthing point 2.4.1 Identifcation of Elements at Risk was used to assess the accuracy of classifcation. The ele- ments at risk identifed for the study areas include cropping The frst step in vulnerability analysis was to identify the land and rural settlements (i.e. homesteads), because other elements at risk in the study area. The elements at risk are land cover classes are not important from food risk point of defned as the level of exposure with reference to agricul- view. It is noted here that image classifcation did not yield tural felds, buildings/infrastructure, population, economic roads as one land use classifcation. Finally, inundation lay- activities, public services and utilities, etc., which can be ers were overlaid on land use layer to obtain the overlaid impacted by the food hazard (Dewan 2013). In this study, zones. From the ArcGIS overlay analysis, diferent sort of elements of risk were identifed by analysing satellite images inundation statistics was generated. in the GIS environment and hence obtaining land use map, followed by overlying the elements onto food inundation maps. Field observations and interviews were conducted to 2.4.2 Assessment of Flood Damage Functions verify the elements identifed. Land use/land cover (LULC) dataset was generated from The quantifcation of vulnerability depends on the suscepti- the digital image classifcation of Landsat, ETM + satellite bility of ‘elements at risk’. It can be termed as the degree of images of 2011, downloaded from Global Land Cover Facil- loss to a given element at a given severity level. It is usually ity (https​://glovi​s.usgs.gov/). Among various image classi- expressed on a scale 0 (no damage) to 1 (total loss) unit. The fcation techniques, the maximum likelihood algorithm is present study considers ‘depth of inundation’ as the main shown to produce useful outcome in deriving LULC (Dewan parameter for assessing food damage functions for crop- and Yamaguchi 2009; Jia 2019). Therefore, in this study lands and rural settlements. The study further considers the supervised classifcation of maximum likelihood algorithm direct economic damages of foods. Considering depth as the was applied to classify Landsat images into discrete land food damage parameter, depth–damage relationships (alter- LULC categories. An overall workfow in preparing LULC natively called loss functions or vulnerability functions) map is shown in Fig. 2. The classifcation was performed on were developed for diferent elements at risk: crops and set- false colour composition of bands 4, 3 and 2 into following tlements. Depth–damage relationship presents information

1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 231 on the relationship of food damage of a certain element to stage-damage function. The following equations were fol- a certain depth of fooding (or stage) (Smith 1994). lowed to estimate expected damage. In this study, for developing depth–damage relationship D = vul × P × A, for crop, the food damage data were collected from diferent secondary literature and organizations, and extensive inter- 1 = × D = × D views with the local people were conducted as part of the ED Probability T , questionnaire survey. A total of 120 local people including farmers, fshermen, and small businessman from the study where D is total direct property damage per cell of the raster site were interviewed as household basis. A questionnaire map, ‘vul’ is the vulnerability value per cell which is the survey was preferred for this interview session as it pro- function of Depth (DP) in meter and duration (DR) of inun- vides insight into the information of inundation depth and dated land in days, A is the area of each cell in sq.m and P is associated food damages. The survey was conducted fol- the property value in monetary terms of each cell. Here, ED lowing a random sampling method to select respondents for is the expected damage and T is the return period of food. the household interviews. The structured questionnaire was Property value data per hectare of each land use class frst pre-tested in 15 randomly selected households. Then, were collected from the feld survey. Data on average unit modifcations were made before the actual interviews of the prices of houses and cropping land under the present cir- sampled households. Additionally, the questionnaire was cumstances were collected from the feld survey. After that, administered to respondents who (i) were aged 20 years a land use-based raster map showing the monetary value for and above, (ii) had lived in the respective area for at least each land parcel was prepared (Economic Value in Fig. 3). 15 years, and (iii) were main decision makers in the house- Direct damages to properties of economic units were classi- hold, and/or, in the absence of a family head, it was made fed as settlement and agricultural damages (Vulnerability in with appropriate representative and knowledgeable member Fig. 3). Then, the expected damage value was classifed into of the household. The questionnaire survey covered the local several defned classes using GIS environment (Expected perceptions on crop and settlement damages associated with Damage in Fig. 3). The single output map algebra in Fig. 3 diferent remarkable food events such as 1988, 1998, 2004 was the product of direct property damage (D) divided by and 2007 as well as collected information on damage corre- return period pf food (T). The conceptual model for entire sponding inundation depths. Damage was assessed in terms food risk mapping is shown in Fig. 3. of the amount of money (presented as percentage of the total production value) necessary to recover the original produc- tion. Based on this food depth–damage information from 3 Results and Discussion questionnaire survey, depth–damage curve was developed. For developing the depth–damage relationship for set- 3.1 Frequency Analysis of Water Levels tlement, a valuation survey was conducted for the settle- ment vulnerability assessment. Following the study by Islam The ftted PDFs and the corresponding values of PPCC for (2005), settlements were classifed into four types such as annual maximum water levels of the selected gauge stations brick foor–brick wall (BB), brick foor–CI sheet wall (BC), and for maximum discharge for Kushiyara River at Sherpur mud foor–CI sheet wall (MC), and mud foor–mud wall (MM). For the selected properties, the survey quantifed the damage of all items due to food and their current value based on type, quality and degree of wear. This included information on the height above the foor of each item or the height taken as standard from house to house. The informa- tion for all samples of each element class was then averaged and stage-damage curves were constructed.

2.5 Flood Risk Assessment

As risk is a combination of hazard, vulnerability and expo- sure (i.e. elements at risk) (Skakun et al. 2014), in the fnal step of risk assessment, the expected damage of the risk element was estimated first by combining replacement values assessed for the elements and then following the Fig. 3 Conceptual model for food risk assessment

1 3 232 M. A. A. Baky et al.

Table 2 Fitted PDFs and the PDF Return period PPCC Rank corresponding values of PPCC for pre-monsoon peak water 2.33 5 10 20 50 100 level (m PWD) data of at Habigang (SW 159) LN2 6.58 7.12 7.50 7.83 8.23 8.50 0.98130 4 LN3 6.59 7.12 7.49 7.80 8.16 8.40 0.98201 3 P3 6.59 7.12 7.49 7.80 8.16 8.40 0.98202 2 LP3 6.62 7.13 7.47 7.76 8.09 8.31 0.98209 1 EV1 6.47 7.04 7.50 7.94 8.51 8.94 0.96622 5

Table 3 Fitted PDFs and PDF Return period PPCC Rank the corresponding values of PPCC for pre-monsoon peak 2.33 5 10 20 50 100 water level (m PWD) data of Kushiyara River at Sherpur (SW LN2 7.49 8.34 8.96 9.51 10.17 10.63 0.97034 4 175.5) LN3 7.55 8.22 8.62 8.93 9.24 9.44 0.99620 1 P3 7.54 8.23 8.64 8.95 9.28 9.48 0.99580 2 LP3 7.71 8.33 8.65 8.86 9.04 9.14 0.99448 3 EV1 7.24 8.07 8.75 9.40 10.24 10.87 0.94093 5

Table 4 Fitted PDFs and the PDF Return period PPCC Rank corresponding values of PPCC for pre-monsoon peak water 2.33 5 10 20 50 100 level (m PWD) data of Surma- Meghna at Markuli (SW 270) LN2 6.58 7.12 7.50 7.83 8.23 8.50 0.98130 4 LN3 6.59 7.12 7.49 7.80 8.16 8.40 0.98201 3 P3 6.59 7.12 7.49 7.80 8.16 8.40 0.98202 2 LP3 6.62 7.13 7.47 7.76 8.09 8.31 0.98209 1 EV1 6.47 7.04 7.50 7.94 8.51 8.94 0.96622 5

Table 5 Fitted PDFs and the PDF Return period PPCC Rank corresponding values of PPCC for pre-monsoon peak water 2.33 5 10 20 50 100 level (m PWD) data of Surma- Meghna at Azmiriganj (SW LN2 5.31 5.91 6.34 6.73 7.20 7.52 0.97644 4 271) LN3 5.49 6.02 6.35 6.61 6.89 7.07 0.98432 2 P3 5.45 6.01 6.38 6.68 7.02 7.24 0.98556 1 LP3 5.35 5.92 6.31 6.64 7.02 7.28 0.98323 3 EV1 5.29 5.91 6.41 6.89 7.52 7.99 0.95344 5

Table 6 Fitted PDFs and the PDF Return PPCC Rank corresponding values of PPCC for pre-monsoon peak discharge 2.33 5 10 20 50 100 data of Kushiyara River at Sherpur (SW_175.5) LN2 1555 1960 2288 2602 3008 3311 0.95961 4 LN3 1628 1944 2147 2312 2494 2611 0.97571 3 P3 1614 1939 2157 2337 2541 2675 0.97718 2 LP3 1607 1970 2222 2433 2670 2824 0.97742 1 EV1 1527 1881 2168 2444 2801 3069 0.95922 5

(SW_175.5) station are shown in Tables 2, 3, 4, 5, and 6. The are shown in Fig. 4. It is seen that the observed values fall probability plots along with 90% confdence interval for the well within the 90% confdence interval of the ftted distribu- annual maximum water levels of the selected gauge stations tions for annual maximum water level.

1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 233

Fig. 4 Probability plot along with 90% confdence interval of the ftted distributions to the pre-monson peak water level data of diferent gauge stations in Baniachong Upazila

3.2 Model Calibration and Validation roughness coefcient ‘n’ in the river. Similar to the study by Dutta and Nakayama (2009), land use types were the After simulating the hydrodynamic model Delft3D, the frst basis for estimating the roughness coefcients for rivers and step was to calibrate and validate the model’s output with surface in this present study. From the calibration, a close observed data. Water level (WL) data for the month April agreement between the observed and calculated water level to May 2007 (pre-monsoon) of Markuli (SW 270) station (Fig. 5) using the roughness coefcients between 0.019 and were used for calibration purpose. The selected Markuli 0.023 was found. (SW_270) station belongs to the Surma- sys- After calibration, the model was validated against the tem (Fig. 1). The calibrated parameter was the Manning’s water levels (WL) for the month April to May 2007 of

1 3 234 M. A. A. Baky et al.

10 Table 7. It was found that overall with the increase in return 9 period the inundated area increases substantially for the 8 D) food class of “Low (0–2 m)” and “Medium (2–4 m)”. How- 7 PW 6 ever, exception is 20-year return period of food, where inun- (m l 5 dated area for the food class of “Low (0–2 m)” decreased ve Observed WL le 4 as compared to the food class of “Low” at 10-year return 3 Computed WL period. Noticeably, at 2.33-year food event, only the north Water 2 side of the study area was inundated, but with greater return 1 period of food the western and the southern sides of the 0 01-Apr-07 14-Apr-07 27-Apr-07 10-May-07 23-May-0705-Jun-07 study area were inundated gradually. There was no inundated Date (Year 2007) area for “High (4–6 m)” class of food at 2.33-year return period; however, at greater reoccurrence of food (greater Fig. 5 Water level calibration of the model at Markuli station (SW than 10-year) inundated area of “High” class of food was 270) present with slight extent (~ 1 km2). The percentage area of inundation increased from 28.77 to 80.28% correspond- ing to 2.33-year return period to 100-year return period, SW 271 respectively, for food class low. Whereas for medium and 7 high class foods, the area of inundation increased from 0.31 6 to 10.23% and 0 to 0.28% for return period 2.33 years to 5 100 years, respectively. PWD)

(m 4 At low reoccurrence interval of food, the major river 3 systems are not always seen to play major role in fooding level 2 the adjacent foodplains by their overbank fows. Rather, Observed WL Water foodplains are inundated from the fows coming from local 1 Computed WL 0 foodplain channels connected with the major river systems 01-Apr-0714-Apr-07 27-Apr-0710-May-07 23-May-0705-Jun-07 (Jeb and Aggarwal 2008; Luo et al. 2018; Tanaka et al. Date (Year 2007) 2017). This is why the distribution of foodplain inunda- tion can be sporadic at low reoccurrence interval of food. Fig. 6 Water level validation of the model at Azmiriganj station (SW However, at later stages of food as water levels continue 271) to rise, foodplains get closer to major river and meet fows directly coming from the river (Fantin-Cruz et al. 2011; Azmiriganj (SW 271) station at Surma-Meghna river. The Karim et al. 2016; Yin et al. 2013; Zin et al. 2018). Then, validation shows that the observed and computed water lev- a vast expanse of the foodplain is inundated. The pattern els are close (Fig. 6). The computed water level at this sta- of inundation in the study site showed similar behaviour tion was found to vary within − 0.17 and the coefcient of at diferent return periods of food. Except few areas, the determination (R2) is 0.72. Further validation was carried whole study site was inundated (Fig. 8) by the overbank out between the observed food extent of 1998, a 100-year fows coming from the major rivers (Fig. 1) (the Kushiyara, food event (Islam and Chowdhury 2002), and the modelled the Khowai, the Surma-Meghna and the Barak) at recurrence 100-year food extent (Fig. 7). The extent of 1998 food was interval of food greater than 20. At relatively low recurrence maximum from 27th July 1998 to 07th September 1998 in intervals of food, a small portion of the study site, northern the study site as per BWDB (2010). RADARSAT image portion mainly, was inundated possibly by local foodplain covering the 1998 inundation extent was found available for channels only. The foods with a relatively low return period 26th August, which is within the specifed range of maxi- have a large infuence on the annual risk. At the same time mum food extent. Therefore, this RADARSAT image was as these foods may cause relatively low economic damage used for delineating the observed inundation extent of 1998 per event, their relatively frequent occurrence means that food. It was found 76.5% accuracy between the modelled they should be fully considered in food risk assessments and the observed inundation extents when a GIS overlay (Ward et al. 2011). operation was performed between these two extents. The inundation extent in the study area simulated by Delft3D model is comparable to other studies. The percent- 3.3 Inundation Maps age of fooded area in Baniachong Upazila was 86.63 as on June 13, 1998 (BWDB 2010). Study conducted by Bhuiyan Simulation by Delft3D model yielded foodplain inundation et al. (2010) termed the 1998 food is a return period of 75 to depths at diferent return periods, as presented in Fig. 8 and 100 years. In this study, for 100-year return period, 80.28%

1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 235

Fig. 7 Map showing 100-year food inundation computed from modelled scenario (left) and 26 August 1998 food inundation from observed scenario

(399.34 km2) area was found under fooding (Table 7), which classes is representative to a typical rural area of Bangladesh is quite close to the BWDB study. (e.g. Khan et al. (2015), Parvin et al. (2017)). Figure 10 presents the percentage of inundation area 3.4 Inundation of Diferent Land Use Categories for each land use class at diferent return periods of food. It is found from Fig. 10 that the afected area increases Supervised classifcation of LANDSAT image with ArcGIS with the increase of return period and food depth for all yielded diferent land cover existing in Baniachong. The land land use classes. It is noticeable that, with the increase in cover map was assessed against Google Earth image. The return periods from 2.33 to 100 inundated areas become assessment was 30 m apart sampling basis, since the LAND- more than tripled for land use classes rural settlement (~ 37 SAT image used in this study is approximately 30 m in reso- to ~ 126 km2), urban settlement (~ 13 to ~ 63 km2) and bare lution. Bai et al. (2015) also used Google Earth images when land (~ 14 to ~ 57 km2), and more than doubled for land assessed land use map of China. Overall accuracy in the Bai use class cropping land (~ 76 to ~ 188 km2). The percent- et al. (2015) study varies from 48.6 to 68.9%. In the present age inundation of urban settlement (~ 16 to ~ 85) is higher study, the assessment shows 70%, 100%, 60%, 62% and 60% than that of rural settlement (~ 15 to ~ 69) with the increase accuracy in water bodies, urban settlement, rural settlement, in return period. Water bodies were inundated much more cropping land and bare land categories, respectively, in the (~ 29% at 2.33-year food and ~ 90% at 100-year food) than derived land cover map. that of any other classes of land use. It refects the loss of The land use map of Baniachong is shown in Fig. 9. capture fsheries during fooding. The rising trend of inun- About ~ 4%, ~ 18%, ~ 27%, ~ 37% and ~ 15% are covered by dated area for cropping land, which is the most dominant water bodies, urban settlement, rural settlement, cropping land use type, decreases with the increase of return periods. land and bare land, respectively. This distribution of land use Noticeably, in case of higher reoccurrence interval of food

1 3 236 M. A. A. Baky et al.

Fig. 8 Inundation depth (food hazard map) for diferent return periods in Baniachong Upazila

Table 7 Depth-wise area inundated at diferent return periods of food structures, such as road network, obstruct lateral and lon-

2 gitudinal connectivity of food water fuxes (Kumar et al. Return Area inundated ­(km ) 2014), thereby reducing the chance of inundation in the period of food Flood depth (m/PWD) settlement areas of Baniachong Upazila. As a result, only Low (0–2) Medium (2–4) High (4–6) a negligible portion of the settlement areas (~ 28% for rural settlement and ~ 16% for urban settlement at 2.33-year 2.33 143.80 (28.77%) 1.55 (0.31%) 0 (0%) food) were inundated at low reoccurrence interval of food 10 312.28 (62.77%) 4.38 (0.88%) 0.08 (0.01%) as found from the simulation. However, at greater reoccur- 20 304.09 (61.18%) 12.57 (2.41%) 0.08 (0.01%) ring interval of food, the settlement areas received food 50 427.30 (85.91%) 21.38 (4.22%) 0.46 (0.09%) water as the food fows defeat the settlement elevation and 100 399.34 (80.28%) 50.87 (10.23%) 1.4 (0.28) or overtop the man-made structures. This led to abrupt inundation of the settlement areas, thereby increasing the inundated area in percentage (~ 28% at 2.33-year food (e.g. 50 and 10-year foods), inundated area for cropping to ~ 94% at 100-year food) (Fig. 10). Cropping or agricul- land remains the same with the increase of return period. ture land areas, which are typically as depressions or low The settlement areas (both urban and rural) in the study elevated zone in active and older foodplain, often receive site are slightly elevated locally, as like as other settlement food water from major rivers as well as from local food- areas in Bangladesh (Choudhury 1973). This is why these plain channels (Charlton 2008). The cropping land area in areas usually do not receive food fow from major riv- Baniachong Upazila most probably has similar character- ers nearby or even from local foodplain channels at low istics in terms of food water connectivity with the major reoccurrence interval of food. Furthermore, man-made rivers and the local foodplain channels. Therefore, most

1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 237

Fig. 9 Land use map of Baniachong Upazila

1.2 100 1

80

(%) 0.8

area 60 Cropping land 0.6

Rural settlement lnerability

40 Vu 0.4 Cropping land Water bodies nundation I 20 Bare land 0.2 Rural settlement Urban settlement 0 0 020406080 100 0 0.511.522.5 3 Return period (years) Depth of water (m)

Fig. 11 Fig. 10 Percent of inundation area for each land use class at diferent Flood damage vulnerability (depth–damage function) of return periods (years) cropping land and rural settlement

3.5 Floodplain Damage Vulnerability of the area of cropping land was inundated at low reoc- currence interval of food (~ 40% and ~ 80% at 2.33 and 3.5.1 Damage Function and Damage Vulnerability 10-year foods, respectively) and unlike settlement areas, Mapping the inundated area did not increase abruptly and even did not change at greater reoccurrence interval of food (from Figure 11 shows the depth–damage curves for two elements 50- year to 100-year food) (Fig. 10). of risks: cropping lands and rural settlements for the study

1 3 238 M. A. A. Baky et al. site. Depth–damage functions for these two elements were 100-year food. These areas are close to the rivers (Fig. 12), constructed with the help of hazard maps shown in Fig. 8. It and further generally laying at low elevations (Elevation is noted here that the damage function shown here for rural map—Fig. 1). On the other hand, ~ 42% of the total cropping settlement refers to an average for four dominant types of land area is “low” to “medium high” vulnerable to 100-year settlements usually found in the study site (as discussed in food. Most of these areas tended to be further away from Sect. 2.4.2). This damage function was used to represent the the high drainage density areas. Signifcantly, the results in physical vulnerability of the rural settlements since it was Fig. 12 depict the fact that the cropping land in northern not possible to distinguish the four diferent types either in portion of the study site is much more vulnerable to food the satellite image processing or through feld survey. than any other area of the study site. This is due to the fact Using the hazard map (Fig. 8) and the stage-damage that the northern portion is very close to the Suriya-Kalakhai curve (Fig. 11), crops vulnerability maps for diferent return River system (Fig. 12), one of the major rivers in the study periods of food were constructed, as shown in Fig. 12. In the site. Furthermore, the area is a depression zone identifed crop vulnerability mapping, the food depths were divided in the elevation map in Fig. 1. Consequently, the extent of into fve scales (0–0.5 m, 0.5–1 m, 1–1.5 m, 1.5–2 m and food damage would be higher in norther portion than any 2 m and above) and their respective vulnerability is: very low other portion of the study site. Note that, this proposition vulnerable (0–0.25), low vulnerable (0.25–0.45), medium is true for same food, in terms of intensity and exceedance high vulnerable (0.45–0.65), high vulnerable (0.65–0.84) probability. However, one might argue with diferent varie- and very high vulnerable (0.84–1) for diferent return peri- ties of crop as crop vulnerability can difer from one variety ods of food. to another variety (Cutter 1996). This is probably a scope of From Fig. 12, it is found that crop vulnerability increases further study in future. with the increase in return period. Almost 20% of the total Rural settlement vulnerability maps for diferent return cropping land area is “high” and or “very high” vulnerable to periods of food were constructed, as shown in Fig. 13. In

Fig. 12 Vulnerability rank for cropping land at diferent return periods of food

1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 239

Fig. 13 Vulnerability rank for rural settlement at diferent return periods of food the rural settlement vulnerability map, the food depths were area at 100-year food). These areas where rural settlement divided into six scales (0–0.5 m, 0.5–1 m, 1–1.5 m, 1.5–2 m, vulnerability with “low” and or “very low” increases are 2–2.5 m and 2.5 m and above) and their respective vulner- tended to be either close to the Suti River and the Barak ability is: very low vulnerable (0–0.15), low vulnerable River (Fig. 13) or laying at relatively low elevations (Eleva- (0.15–0.35), medium low vulnerable (0.35–0.6), medium tion map—Fig. 1). As a whole, distance from river to set- high vulnerable (0.6–0.75), high vulnerable (0.75–0.9) and tlement determines the settlement vulnerability in the study very high vulnerable (0.9–1) for diferent return periods. site. However, type of settlement is obviously a factor in this The result of vulnerability map for rural settlement show- regard, but it is out of scope in the present study. ing in Fig. 13 depicts that rural settlement vulnerability increases with the increase in return period. Overall, irre- 3.5.2 Damage estimation and risk mapping spective of any class of vulnerability, ~ 28% of the total rural settlement area is vulnerable to food at 2.33-year return The expected damage of the inundated land use types was period. This percentage becomes ~ 95% at 100-year food. estimated using equation outlined in Sect. 2.5. The value of Now, if looking on the basis of vulnerability class, only 0.2% P (property value) was found to be Tk. 9.4 (Tk. 2820 per of the total rural settlement area is “high” and or “very high” cell) for agriculture and Tk. 590 for settlement (Tk. 177,000 vulnerable to food at 2.33-year return period. However, at per cell). Thus, raster-based damage maps for various return 100-year food this percentage increases to ~ 8. This increas- of foods were produced. ing trend is probably due to the nearby Kalni River (Fig. 13). Figure 14 shows the expected damage or risk map for Noticeably, rural settlement area with “low” and or “very cropping land at diferent return periods of food. Figure 15 low” vulnerable to food increases sharply with the increase presents the percentage of cropping land area with difer- in return period (~ 15% of the total rural settlement area at ent levels of risk at diferent return periods of food. From 2.33-year food increase to ~ 30% of the total rural settlement Figs. 14 and 15, it is found that overall the cropping land

1 3 240 M. A. A. Baky et al.

Fig. 14 Expected damage or risk map for cropping land at diferent return periods of food

50 but in northern portion it is always under risk at all level of 40 foods (Fig. 14).

land area The reason why cropping land areas with “Low” and 30 “High” food risk decrease with the increase in return period 20 cropping is probably because these areas are shifted to either risk class

total 10 of “Medium” or “Very high” when afected by greater reoc- currence interval of food. Overall, existing topography and % of 0 020406080100 river-channel network performing as vulnerability element Return period (years) and food depth as hazard element determine the spatial dis-

Very low Low Medium High Very high tribution of diferent levels of food risk for the cropping land in the study site. The northern portion of the study area has depression (Elevation map—Fig. 1), and this is probably Fig. 15 Percent of cropping land area with diferent levels of risk at diferent return periods of food the main reason for which the area is under food risk at all level of foods. Furthermore, the Kalni River at north is also responsible for the area to be fooded at all level of foods. area with diferent classes of food risk increases with the The drainage density in southern portion of the study area increase in return period. However, cropping land areas with is relatively higher (Fig. 14), thus increasing the food risk “Low” and “High” food risk decrease (~ 30 km2 to ~ 28 km2 level for cropping land in respective area. for “Low” and ~ 28 km2 to ~ 12 km2 for “High”) with the Figure 16 shows the risk maps for rural settlement at dif- increase in return period from 10 to 100-years and 50 to 100- ferent return periods of food. Figure 17 presents the per- years, respectively (Fig. 14). The cropping land in southern centage of rural settlement area with diferent levels of risk portion of the study area is not under risk at 2.33-year food, at diferent return periods of food. Figures 16 and 17 show

1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 241

Fig. 16 Expected damage or risk map for settlement at diferent return periods of food

60 food. However, at greater reoccurrence interval of foods, 50 these areas, particularly the areas close to the Suti River, fall 40 under the risk classes from “Medium low” to “Very high”.

settlement area 30 The reason could be the existence of intricated network of 20 rural river-channels close to the area as shown in Fig. 16. 10

total Overall, the study fnds that areas with diferent levels of 0 food risk do not correspond to the areas of inundation at dif- % of 020406080100 Return period (years) ferent return periods of food. While the areas of inundation increase with increase in return period of food, the areas Very low Low Medium Low with diferent levels of food risk are determined by depth Medium high High Very high of inundation and depth–damage function (i.e. vulnerability index). Fig. 17 Percent of rural settlement area with diferent levels of risk at diferent return periods of food 4 Conclusions that except the area covered by the risk classes of “Low” and “Medium high” all other classes increase with the increase Baniachong Upazila represents a fash food area where Boro in return period. Signifcantly, the area covered by the risk rice is the dominant crop, and this variety of crop is a very class of “Very low” increases much more (~ 14 km2 at 2.33- important element for estimating risk. Area of inundation year food to ~ 67 km2 at 100-year food) than any other depth increases substantially with increasing return period, classes. The rural settlement areas located south and south- which has a considerable impact on the area of cropland and east of the Suti River (Fig. 16) are not at risk to 2.33-year also the area of settlements. It was found that cropland is

1 3 242 M. A. A. Baky et al. highly vulnerable at 2.8-m depth, while settlement is highly Open Access This article is distributed under the terms of the Crea- vulnerable above 3-m depth. tive Commons Attribution 4.0 International License (http://creat​iveco​ mmons.org/licen​ ses/by/4.0/​ ), which permits unrestricted use, distribu- Traditionally, hydrologic/hydraulic models are commonly tion, and reproduction in any medium, provided you give appropriate used to study or delineate the potential areas of food hazard credit to the original author(s) and the source, provide a link to the at given recurrence interval of food. Determining the food Creative Commons license, and indicate if changes were made. hazard using one of the popular hydraulic models, and then delineate the food vulnerable and risk areas as carried out in this study for Baniachong Upazila is important for decision References makers, planner and overall management activities. Impor- tantly, this study shows an efective approach of integrating Aff Z, Chu H-J, Kuo Y-L, Hsu Y-C, Wong H-K, Ali MZ (2019) Residential food loss assessment and risk mapping from high- GIS, hydraulic model and feld survey in the study of food resolution simulation. Water 11:751 risk assessment, and this approach successfully assesses Alexander D (1991) Natural disasters: a framework for research and food risk in the study site, where risk as a concept has been teaching. Disasters 15:209–226 placed in terms of theoretical framework. Furthermore, the Bai Y, Feng M, Jiang H, Wang J, Liu Y (2015) Validation of land cover maps in China using a sampling-based labeling approach Remote study showing approach to food risk assessment is relatively Sensing 7:10589–10606 inexpensive and easy to manage and more signifcantly Baky AA, Zaman AM, Khan AU (2012) Managing food fows for crop allows interactive use by food managers for continuing production risk management with hydraulic and GIS modeling: improvement. case study of agricultural areas in Shariatpur. APCBEE Procedia 1:318–324 Nonetheless, the study has a number of limitations mainly Baugh CA, Bates PD, Schumann G, Trigg MA (2013) SRTM vegeta- due to resource and time constraints. Damage due to foods tion removal and hydrodynamic modeling accuracy. Water Resour can depend on a number of factors, including depth of food Res 49:5276–5289 inundation, duration of fooding, fow velocity, timing of Bhuiyan SR, Baky AA (2014) Digital elevation based food hazard and vulnerability study at various return periods in Sirajganj Sadar occurrence, rate of rise of food, etc. However, the study did Upazila. Bangladesh Int J Disaster Risk Reduct 10:48–58 not consider any other potential factor except ‘depth of inun- Bhuiyan MJAN, Dutta D (2012) Analysis of food vulnerability and dation’ as the parameter for assessing food damage func- assessment of the impacts in coastal zones of Bangladesh due to tions. While fow velocity is typically an important damage potential sea-level rise. Nat Hazards 61:729–743 Bhuiyan JA, Jakobsen F, Khan AS, Bhuiyan S (2010) Flood charac- parameter in such fash food area, not considering it may teristics of Upper Meghna River Basin, Bangladesh. Int J Model be justifed, since the area is much inland corresponding Simul 30:428–433 to a higher order catchment and the fact that the efect of Bhuiyan SR, Baky AA, Hoque MM, Salehin M, Islam AS, Islam T velocity is maximum in the hilly, the border areas, which (2014) Development of cyclone hazard maps with efect of cli- mate change scenario for Barguna Sadar Upazila. Bangladesh Int gradually diminishes with the increase in the order of the J Surf Groundwater Manag 01:6 catchment. However, velocity is one important parameter Brammer H (1999) Agricultural disaster management in Bangladesh. while assessing food damage due to storm surges and future University Press Ltd, Dhaka study should be undertaken considering this parameter. Burrel BC, Davar K, Hughes R (2007) A review of food management considering the impacts of climate change. Water Int 32:342–359 Another limitation is that only the direct economic dam- BWDB (2010) Flood report 1998. Bangladesh Water Development ages of foods has been considered, while some studies (i.e. Board, Ministry of Water Resources Smith (1994)) indicated that indirect food damage (with Charlton R (2008) Fundamentals of fuvial geomorphology. Routledge, multiplier efect) may assume a signifcant proportion of the London Choudhury JR (1973) Low-rise low-cost housing and extreme wind total food damage. While vulnerability due to food hazards related problems in Bangladesh National Bureau of Standards. encompasses physical, social, economic and environmen- US Department of Commerce, Washington DC tal dimensions, only the physical vulnerability aspects have Chowdhury NT (2010) Water management in Bangladesh: an analytical been considered in the present study because of resource review. Water Policy 12:32–51 Chowdhury J, Karim M (1996) A risk-based zoning of storm surge limitation and lack of time to carry out the investigation. prone area of the Ganges Tidal Plain Journal of Civil Engineering. Inst Eng Bangladesh 24:221–233 Acknowledgements The study was fnancially supported by the Cli- Chowdhury J, Rahman M, Salehin M (1997) Flood control in a food- mate Change Trust Fund of Climate Change Cell under Ministry of plain country: experiences of Bangladesh, Institute of Flood Con- Forestry and Environment (MoEF) of Government of Bangladesh. It trol and Drainage Research Publication of the Islamic Education, was being implemented by Climate Change Study Cell housed at Insti- Scientifc and Cultural Organization (ISESCO), Rabat, Morocco tute of Water and Flood Management (IWFM), Bangladesh University Crichton D (1999) The risk triangle. Nat Disaster Manag 102:103 of Engineering and Technology (BUET). The authors of the study sin- Cunnane C (1978) Unbiased plotting positions—a review. Journal of cerely acknowledge the help from Dr Ashraf Dewan, Senior Lecturer, Hydrology 37:205–222 School of Earth and Planetary Sciences (EPS) for providing a general Cutter SL (1996) Vulnerability to environmental hazards. Prog Hum outline to prepare this study. He is also acknowledged for providing the Geogr 20:529–539 RADARSAT image of 1998 food event, which was used in this study for validation purpose of the hydraulic model’s result.

1 3 Flood Hazard, Vulnerability and Risk Assessment for Diferent Land Use Classes Using a Flow… 243

Deltares DD-F (2014) Users manual. Simulation of multi‐dimen- Khan M, Sabbir M (2018) Flood hazard mapping of food- sional hydrodynamic fows and transport phenomena, including plain using HEC-RAS 1D/2D coupled model sediments Khan MMH, Bryceson I, Kolivras KN, Faruque F, Rahman MM, Demir V, Kisi O (2016) Flood hazard mapping by using geographic Haque U (2015) Natural disasters and land-use/land-cover change information system and hydraulic model: Mert River, Samsun. in the southwest coastal areas of Bangladesh. Reg Environ Change Turkey. Adv Meteorol 2016:9 15:241–250 Dewan AM (2013) Modeling food hazards. In: Dewan AM (ed) Floods Kumar R, Jain V, Babu GP, Sinha R (2014) Connectivity structure of in a megacity: geospatial techniques in assessing hazards, risk and the Kosi megafan and role of rail-road transport network. Geo- vulnerability. Springer, Berlin, pp 129–137 morphology 227:73–86 Dewan AM, Yamaguchi Y (2009) Land use and land cover change Luo P et al (2018) Flood inundation assessment for the Hanoi Central in Greater Dhaka, Bangladesh: using remote sensing to promote Area, Vietnam under historical and extreme rainfall conditions. sustainable urbanization. Appl Geogr 29:390–401 Sci Rep 8:12623 Dewan AM, Islam MM, Kumamoto T, Nishigaki M (2007) Evalu- Masood M, Takeuchi K (2012a) Assessment of food hazard, vulner- ating food hazard for land-use planning in Greater Dhaka of ability and risk of mid-eastern Dhaka using DEM and 1D hydro- Bangladesh using remote sensing and GIS techniques. Water dynamic model. Nat Hazards 61:757–770 Resour Manag 21:1601–1612 Masood M, Takeuchi K (2012b) Flood Hazard and Risk Assessment Dutta D, Nakayama K (2009) Efects of spatial grid resolution on in Mideastern part of Dhaka, Bangladesh FLOOD river fow and surface inundation simulation by physically Merz B, Thieken A, Gocht M (2007) Flood risk mapping at the local based distributed modelling approach. Hydrol Process Int J scale: concepts and challenges. In: Begum S, Stive MJF, Hall 23:534–545 JW (eds) Flood risk management in Europe. Springer, Berlin, pp Fantin-Cruz I, Pedrollo O, Castro NM, Girard P, Zeilhofer P, Ham- 231–251 ilton SK (2011) Historical reconstruction of foodplain inunda- Mirza MMQ (2011) Climate change, fooding in South Asia and impli- tion in the Pantanal (Brazil) using neural networks. J Hydrol cations. Region Environ Change 11:95–107 399:376–384 Parvin GA, Ali MH, Fujita K, Abedin MA, Habiba U, Shaw R (2017) Filliben JJ (1975) The probability plot correlation coefcient test for Land use change in southwestern coastal Bangladesh: Conse- normality. Technometrics 17:111–117 quence to food and water supply. In: Banba M, Shaw R (eds) Giustarini L, Chini M, Hostache R, Pappenberger F, Matgen P (2015) Land use management in disaster risk reduction. Springer, Berlin, Flood hazard mapping combining hydrodynamic modeling and pp 381–401 multi annual remote sensing data. Remote Sens 7:14200–14226 Patro S, Chatterjee C, Mohanty S, Singh R, Raghuwanshi N (2009) Hasan SI (2006) Application of geoinformatics for food study at Flood inundation modeling using MIKE FLOOD and remote sens- Tarapur Union of Gaibandha. M. Sc. Thesis, IWFM, Bangladesh ing data. J Indian Soc Remote Sens 37:107–118 University of Engineering and Technology Paul BK (1995) Farmers’ responses to the food action plan (FAP) of Hazarika N, Barman D, Das A, Sarma A, Borah S (2018) Assess- Bangladesh: an empirical study. World Dev 23:299–309 ing and mapping food hazard, vulnerability and risk in the Paul BK (1997) Flood research in Bangladesh in retrospect and pros- Upper valley using stakeholders’ knowl- pect: a review. Geoforum 28:121–131 edge and multicriteria evaluation (MCE). J Flood Risk Manag Paul SK, Hossain MN (2013) People’s perception about food disaster 11:S700–S716 management in Bangladesh: a case study on the Chalan Area. Hoque R, Nakayama D, Matsuyama H, Matsumoto J (2011) Flood Stamford J Environ Hum Habitat 2:72–86 monitoring, mapping and assessing capabilities using RADAR- Paul SK, Routray JK (2010) Flood proneness and coping strategies: the SAT remote sensing, GIS and ground data for Bangladesh. Nat experiences of two villages in Bangladesh. Disasters 34:489–508 Hazards 57:525–548 Pumchawsaun P (2018) Integrated hydrodynamic and socio-eco- Hoque MA-A, Tasfa S, Ahmed N, Pradhan B (2019) Assessing spatial nomic damage modelling for assessment of food risk in large- food vulnerability at Kalapara Upazila in Bangladesh using an scale basin: the case study of Lower Chao Phraya River Basin analytic hierarchy process. Sensors 19:1302 in Thailand Hunting (1992) FCD/I Agricultural Study, FAP-12, Main Report 1 Rahman R, Salehin M (2013) Flood risks and reduction approaches in Islam KMN (2005) Flood loss potentials in non-agricultural sectors: Bangladesh. In: Shaw R, Mallick F, Islam A (eds) Disaster risk assessment methods and standard loss database for Bangladesh reduction approaches in Bangladesh. Springer, Berlin, pp 65–90 Dhaka. Palok Publishers, Bangladesh Rakib MR, Islam MN, Islam MN (2017) Flood vulnerability mapping Islam A, Chowdhury JU (2002) Hydrological characteristics of the to riverine foods: a study on the old Brahmaputra River. Am J 1998 Flood in major rivers. In: Ali AM, Seraj SM, Ahmad S (eds) Geosci 7:47–58 Engineering Concerns of Flood, pp 227–239 Rodriguez E, Morris CS, Belz JE (2006) A global assessment of the Islam MM, Sado K (2000) Development of food hazard maps of SRTM performance. Photogramm Eng Remote Sens 72:249–260 Bangladesh using NOAA-AVHRR images with GIS. Hydrol Sci Sanyal J, Lu X (2009) Ideal location for food shelter: a geographic J 45:337–355 information system approach. J Flood Risk Manag 2:262–271 Islam A, Bala S, Haque M (2010) Flood inundation map of Bangladesh Shaw R (2006) Critical issues of community based food mitigation: using MODIS time-series images. J Flood Risk Manag 3:210–222 examples from Bangladesh and Vietnam. Sci Cult 72:62 Jeb DN, Aggarwal S (2008) Flood inundation hazard modeling of the Skakun S, Kussul N, Shelestov A, Kussul O (2014) Flood hazard and River Kaduna using remote sensing and geographic information food risk assessment using a time series of satellite images: a case systems. J Appl Sci Res 4:1822–1833 study in Namibia. Risk Anal 34:1521–1537 Jia K et al (2019) Land use and land cover classifcation using Chinese Smith DI (1994) Flood damage estimation-A review of urban stage- GF-2 multispectral data in a region of the North China Plain. damage curves and loss functions. Water Sa 20:231–238 Front Earth Sci 13:327–335 Sohel MSI, Ullah MH (2012) Ecohydrology: a framework for over- Karim F, Petheram C, Marvanek S, Ticehurst C, Wallace J, Hasan M coming the environmental impacts of shrimp aquaculture on the (2016) Impact of climate change on foodplain inundation and coastal zone of Bangladesh. Ocean Coast Manag 63:67–78 hydrological connectivity between wetlands and rivers in a tropi- Stedinger JR (1993) Frequency analysis of extreme events in handbook cal river catchment. Hydrol Process 30:1574–1593 of hydrology

1 3 244 M. A. A. Baky et al.

Tanaka T, Tachikawa Y, Ichikawa Y, Yorozu K (2017) Impact assess- Ward PJ, Moel Hd, Aerts J (2011) How are food risk estimates afected ment of upstream fooding on extreme food frequency analysis by by the choice of return-periods? incorporating a food-inundation model for food risk assessment. Yin J, Yu D, Yin Z, Wang J, Xu S (2013) Multiple scenario analyses of J Hydrol 554:370–382 Huangpu River fooding using a 1D/2D coupled food inundation Tingsanchali T, Karim MF (2005) Flood hazard and risk analysis model. Nat Hazards 66:577–589 in the southwest region of Bangladesh. Hydrol Process Int J Younus M (2012) Autonomous crop adaptation processes to extreme 19:2055–2069 foods: A case study in Bangladesh. In: Water and climate: policy Tran P, Shaw R, Chantry G, Norton J (2009) GIS and local knowledge implementation challenges; Proceedings of the 2nd practical in disaster management: a case study of food risk mapping in Viet responses to climate change conference, 2012. Engineers Aus- Nam. Disasters 33:152–169 tralia, p 124 Tyrna B, Assmann A, Fritsch K, Johann G (2018) Large-scale high- Younus MAF, Harvey N (2014) Economic consequences of failed resolution pluvial food hazard mapping using the raster-based autonomous adaptation to extreme foods: a case study from Bang- hydrodynamic two-dimensional model FloodAreaHPC. J Flood ladesh. Local Economy 29:22–37 Risk Manag 11:S1024–S1037 Zin WW, Kawasaki A, Takeuchi W, San ZMLT, Htun KZ, Aye TH, Van Alphen J, Lodder Q (2006) Integrated food management: experi- Win S (2018) Flood Hazard Assessment of Bago River Basin. ences of 13 countries with their implementation and day-to-day Myanmar J Disaster Res 13:14–21 management. Irrig Drain 55:159–171 Vojtek M, Vojteková J (2016) Flood hazard and food risk assessment at the local spatial scale: a case study Geomatics. Nat Hazards Risk 7:1973–1992

1 3